Models of Arctic-alpine refugia highlight importance of climate and local topography
详细信息 查看全文
- 作者：Annina Niskanen ; Miska Luoto ; Henry Väre ; Risto K. Heikkinen
- 关键词：Generalized boosted model ; GBM ; Spatial modelling ; Species distribution models ; Refugium ; High ; latitude environments
- 刊名：Polar Biology
- 出版年：2017
- 出版时间：March 2017
- 年：2017
- 卷：40
- 期：3
- 页码：489-502
- 全文大小：
- 刊物类别：Biomedical and Life Sciences
- 刊物主题：Ecology; Oceanography; Microbiology; Plant Sciences; Zoology;
- 出版者：Springer Berlin Heidelberg
- ISSN：1432-2056
- 卷排序：40
文摘
Projected climatic warming calls for increased attention to the identification of suitable refugia for the preservation of biota and ecosystems in changing high-latitude environments. One such way is the development of models for drivers of refugia. Here, we investigate the distribution and species richness of Arctic-alpine vascular plant species’ refugia. The study is carried out in an environmentally variable area in N Europe, encompassing the northern boreal to the Arctic-alpine zone. We defined refugia as isolated 1 km × 1 km grid cells with multiple Arctic-alpine plant species occurrences outside their main distribution area and assessed the main environmental factors underlying their distribution and richness using cross-validated boosted regression tree modelling. In the modelling, we examined the effects of climatic, topographic, and geologic factors, and the connectivity of sites with refugia incrementally, i.e. first modelling climatic impact alone, then with separate additions of topographic, geologic and connectivity variables, concluding with a model including all predictor variables. The inclusion of slope and connectivity significantly improved model performance. Although climate has a central role in controlling the occurrence of refugia, topography provides important clues for recognizing heterogeneous locations that harbour refugia with suitable local thermal and moisture conditions. Results suggest considering refugia as, on the one hand, isolated pockets of suitable habitat, but on the other hand as potentially interconnected habitat networks. In general, our study demonstrates that the spatial patterns of refugia can be successfully modelled, but emphasizes a need for high-quality data sampled at resolutions reflecting significant environmental gradients.